Module Details

The information contained in this module specification was correct at the time of publication but may be subject to change, either during the session because of unforeseen circumstances, or following review of the module at the end of the session. Queries about the module should be directed to the member of staff with responsibility for the module.
Title Machine Learning and BioInspired Optimisation
Code COMP532
Coordinator Prof K Tuyls
Computer Science
K.Tuyls@liverpool.ac.uk
Year CATS Level Semester CATS Value
Session 2017-18 Level 7 FHEQ Second Semester 15

Aims

In this module we focus on learning agents that interact with an initially unknown world. Since the world is dynamic this module will put strong emphasis on learning to deal with sequential data unlike many other machine learning courses. The aims can be summarised as:

  1. To introduce and give an overview to state of the art bio-inspired self-adapting methods. 
  2. To enable students to not only learn to build models with reactive input/output mappings but also build computer programs that sense and perceive their environment, pl an, and make optimal decisions. 
  3. To familiarise students  with multi-agent reinforcement learning, swarm intelligence, deep neural networks, evolutionary game theory, artificial immune systems and DNA computing.
  4. To demonstrate principles of bio-inspired methods, provide indicative examples, develop problem-solving abilities and provide students with experience to apply the learnt methods in real-world problems.

Learning Outcomes

A systematic understanding of bio-inspired algorithms that can be used for autonomous agent design and complex optimisation problems.

In depth insight in  the mathematics of biologically inspired machine learning and optimisation methods.

A comprehensive understanding of the benefits and drawbacks of the various methods.

Demonstrate knowledge of using the methods in real-world applications (e.g. logistic problems).

Practical assignments will lead to hands on experience using tools as well as coding of own algorithms.


Syllabus

This module will cover the following topics:

  • Introduction to parallel problem solving from nature/overview (2 lectures)
  • Reinforcement Learning/multi-agent reinforcement learning/replicator dynamics (8 lectures)
  • Swarm Intelligence: Ant System, Ant Colony Optimization/Bee System/Swarm Robotics (6 lectures)
  • Deep Learning: Restricted Boltzman Machines/auto-encoder networks /deep belief networks (8 lectures)
  • Artificial immune systems (4 lectures)
  • DNA computing (2 lectures)

Lecture slides and reading material will be made available to the students.


Teaching and Learning Strategies

lectures - students will be expected to attend three hours of formal lectures in a typical week

tutorials - one hour of weekly seminar given by students in groups, or one hour of tutorial by instructor.


Teaching Schedule

  Lectures Seminars Tutorials Lab Practicals Fieldwork Placement Other TOTAL
Study Hours           30
students will be expected to attend three hours of formal lectures in a typical week
10
one hour of weekly seminar given by students in groups, or one hour of tutorial by instructor.
40
Timetable (if known)              
Private Study 110
TOTAL HOURS 150

Assessment

EXAM Duration Timing
(Semester)
% of
final
mark
Resit/resubmission
opportunity
Penalty for late
submission
Notes
Unseen Written Exam  180  75  Yes  Standard UoL penalty applies  written examination Notes (applying to all assessments) The first report will be due in week 6 and the second report will be due in week 10. The first report will concern a task related to the state of the art literature in RL, evolutionary game theory, swarm intelligence (with a max of 5 pages). The report of the 2nd task will revolve around a student presentation during the tutorial sessions on one of the bio-inspired methods discussed during formal lectures (with a max of 5 pages). 
CONTINUOUS Duration Timing
(Semester)
% of
final
mark
Resit/resubmission
opportunity
Penalty for late
submission
Notes
Coursework  max 5 pages  10  Yes  Standard UoL penalty applies  report  
Coursework  max 5 pages  15  Yes  Standard UoL penalty applies  report 

Recommended Texts

Reading lists are managed at readinglists.liverpool.ac.uk. Click here to access the reading lists for this module.
Explanation of Reading List: